Biomedical Engineering Reference
In-Depth Information
Table 8.2:
Numerical evaluation of the multimodal registration method on
simulated data. The overlapping measures (specificity, sensitivity, and total
performance) are computed after rigid registration and at each grid level of the
non-rigid registration process
Registration
Overlap measure
Grey matter
White matter
Rigid
sensibility
74
7%
76
6%
.
.
specificity
93
.
0%
92
.
8%
Total
performance
87
.
0%
87
.
6%
Non-rigid
sensibility
84
.
7%
86
.
0%
grid level
7
specificity
97
.
2%
96
.
2%
Total
performance
93
.
2%
92
.
9%
Non-rigid
sensibility
86
.
6%
86
.
8%
grid level
6
specificity
98
.
5%
97
.
3%
Total
performance
94
.
6%
93
.
9%
Non-rigid
sensibility
87
.
5%
87
.
0%
grid level
5
specificity
98
9%
98
0%
.
.
Total
performance
95
8%
95
3%
.
.
Figure 8.22: Results of clinical data. Top: results of the rigid registration. The
multiple artefacts are visible: distortions, signal saturation, signal drops. Bottom:
Results of the non-rigid registration. The registration is more accurate, in par-
ticular, for the ventricles and for the cyst. The data are courtesy of “laboratoire
IDM, Hopital de Pontchaillou, Rennes”.
